—

Only this season's games count

Every team starts the season at zero and moves only on the games it plays. Polls, preseason expectations, last season and recruiting have no effect.

Strength and ranking are separate

Each team gets two things:

• Strength is how good a team is. It comes from scores only.
• Ranking is what a team has accomplished. It comes from wins and losses only.

A team's margin of victory never counts toward its own ranking.

How strength works

For each game, the model compares the final margin with the margin the two teams' ratings predicted. Win by more than predicted and your rating goes up. Win by less and it goes down, even though you won.

Big margins stop counting after a point. A 28-point win counts about the same as a 21-point win, and a 60-point win adds almost nothing more. Beating a weak team by the expected amount doesn't change your rating.

Strength is measured in points against an average FBS team, which is 0. Ratings are closer together than real scores, because blowouts are capped and teams are pulled toward the middle until they have played enough games. A 10-point gap is large: the stronger team wins about 80% of the time. That conversion is based on how late-season games from 2018 to 2025 turned out.

How the ranking works

The ranking sorts teams by repeatedly splitting them into a top half and a bottom half:

1. Start with a group of teams (at first, all of FBS) and imagine a team exactly as strong as the group's average.
2. Grade each team's schedule against that average team. A win is worth 1 minus the chance the average team would have won that same game at that site. A loss costs the chance the average team would have won it.
3. Sort the group by grade and split it in half. Teams near the cut (10% of the group on each side) go into both halves.
4. Repeat inside each half until every group is one team. A team that landed in both halves gets the average of its two positions.

This means teams near the top are compared against strong benchmarks, and teams in the middle against average ones. There's no single score behind the order. Each team's game log shows its rank after every week.

In practice: easy wins earn little, hard wins earn the most, losses to strong teams cost little, and losses to weak teams cost nearly a full game.

Margin of victory affects the ranking only through opponents' strength ratings, which is how the model knows how hard each game was. A team's own rating isn't used in its own grade, so an unbeaten team with an easy schedule can't rank higher just because its rating is high.

Strength of schedule

All ratings are calculated together, so your rating depends on your opponents' ratings, and theirs depend on their opponents'. If a team you beat keeps winning, that win becomes worth more over the season. If they start losing, it becomes worth less. There's no separate schedule adjustment.

Why one 5-0 team can rank below another

The model doesn't trust short records. Five wins over weak teams is less evidence than five wins over good teams, and teams are pulled toward the middle until they show otherwise. That's also why no team is far ahead of the pack in September.

FCS and smaller schools

Every team in the results, down to Division III, gets its own rating from its own games. Beating a strong FCS team like North Dakota State is worth more than beating a weak one.

Picking a week

When you pick a week on the timeline, you see the rankings using only games played through that week. Earlier games are re-evaluated with everything known by then, so a week 2 win over a team that turned out to be good looks better in week 8 than it did in week 2.

Early in the season

With only a few games per team, ratings in the middle of the pack are very close together. In September, treat a #6 and a #16 as roughly even until conference play adds more games.

Other details

• Home field is worth 2.5 points. Neutral-site games get no adjustment.
• Bowl and playoff games aren't included, so you can compare the final regular-season ranking with who won the title.
• "Expected win" in a team's game log is the chance of that result based on what every team has shown through the selected week. A low number on a win means a good win. A high number on a loss means a bad loss.
• The conference view uses the same method, treating each conference as one team and using only games between conferences.